(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "A Packet Format Proposal"

INWG #48 
NIC 2] ,596 
A PACKET FORNAT PROPOSAL 
24 January 1974 
V. Crf (SU-DSL) 
Following the INWG working meeting in Hawaii,.L. Pouzin, V. Cerf, 
and C. Sunshine reviewed IG Note #42, and a particularized 
version of the packet format was developed, based on work done 
earlier in September. The result is attached and is proposed as 
the basis or the 1974-75 experiments in inter-networking. The 
format will accommodate both the protocol proposals in INWG 39 
and in INWG 43. Unless there are some strong and cogent objections 
to this format (by the end of February 1974), I hereby recommend 
that it be adopted for our experiments over the near term. Note 
that adoorion at this time does not commit anyone to a 'standard" 
except for the duration of our early experiments. 
Received at NIC 29-JAN-74 
------------------------------<page break>-----------------------------
DEST.: DR.'VINT CERF - STANFORD UNIV. 
RESEAU CYCLADES 
IDEAr IFICATI ON 
EXPERIStENTAL COMMUNICATION PROTOCOL. 
L. PO5% IN 
SCH 515 
Oct. 1973 
revised: Jan. 1974 
DRAFT FOR COSLMENTS 
BASIC IESSAGE FRAME. 
FORWARD 
End to end communication protocols have been discussed at an 
INWG meeting in Brighton on 15 September 1973. Another meeting took 
place at NPL o'! 19 September 1973, in which participants' were: 
V. Cerf, D. Davies, L. Pouzin, R. Scantlebury, P. Wilkinson. This 
note contains some of the conclusions reached at this last meeting. 
1 - BASIC COMMUNICATIONS 
As a stepping stone for building'specific higher level protocols, 
there is a. need for a very straightforward message transfer protocol. 
It is based on the following assumptions: 
'Messages received are copies of messages sent. Every single 
message is delivered as a single piece, not in fragments. (Provision is 
made for fragmentation experiments, however). 
 Messages may be delivered out of order. 
Only core to core format is specified. Synchro, transparency, 
and checksum bits are dependent upon local network interface. 
This basic protocol is not intended to be used as such. It is no more 
than a shell providing a standard way of getting information through an 
arbitrary medium capable of switching complete messages. Typically it 
Should act as a carrier for any kind of end to end protocol resulting from 
user acceptance. 
2 - MESSAGE FRAIE 
SOURCE ADDRESS 24 BITS 
DESTINATION ADDRESS 24 
SPARE 24 
TEXT LENGTH (OCTETS) 16 (MAX. 65,525) 
TEXT 0-524,200 
This whole message may have to be carried as text within some 
network which would wrap it into its local format. Consequently, the 
total length may not exceed 65,536 octets. Since the header iB 11 
octets long, the text length may not exceed 65,525 octets. 
1/3 
------------------------------<page break>-----------------------------
 SCH 515  
3. MESSAGE PROTOCOL 
In order to build a reliable data transmission scheme, some 
protocol is needed to control the message flow. This is left for 
further study. 
An experimental protocol ba'sed on (1) will be worked out. 
It allows fragmentation in GATEWaYs, with reassembly at the final 
destination. 
4. REVISED FORST 
There was not enough time during the September 19 meeting to coin 
a format along' lines suitable for an international standard. Having 
now completed a proposal for a trans-network message format (2), it 
is possible to suggest a revised lay out, without interfering with the 
principles agreed upon. The format below particularizes the general 
format proposed in (2) and would be used for initial experiments. 
USED BY LOCAL NEVORK 2 Bits XX 
HEADER TYPE 2 11 EXPERI 5NTAL 
HEADER LENGTH (-8) 4 010 13 OCTETS 
TEXT LENGTH 16 ........ M. 242* 
LOCAL NETWORK CONTRO 3 XXX 
ECHO 1 0 
RESERVED 12 
MESSAGE IDENTIFICATION 16' 
1 = ECHO 
DESTIN FORT 4 
.. ADDRESS 4 
SOURCE FORST 4 
.. ADDRESS 4 
00Ol 
1010 ARPANET 
,1011 ULICS 
0001 /1100 CYCLADES 
--1101 NPL ' 
LOCAL DESTIN ADDRESS 16 ARPA HOST 
.. SOLACE .. 16 CYCLADES STATe 
This format does not contain any more information than the original 
one, (except for the echo). But using it would simultaneously allow 
another' experiment viz. switch both local and foreign messages without 
reformatting headers. The idea is to assess the proposed international 
format in. view of a possible network standardization. Figure 1 gives an 
alternative presentation of the proposed format. 
Comments would be welcomed. 
5.' REFERENCES 
1. Cerf V., Kahn, R. - Towards Protocols for Internetwork Communication. 
NIC 18764, (SEP 73), 34 p. (INWG 39) 
2. Pouzin, L. - Inter-Connection of Packet Switching Networks. Reseau 
Cyclades, SCH 513, (Oct. 73), 19 p (INWG 42) 
* Initially we agreed to limit message text to 242 octets plus 13 octets of 
header, however, we do not rule out fragmentation experiments with 
larger message lengths. 
2/3 
------------------------------<page break>-----------------------------
o 
(D (D '0 
 0 
o 
o (p 
   o 
o o .: 
o 
o 
o 
(p 
o 
 o 
o 
------------------------------<page break>-----------------------------